Introduction: The Super-efficient USB Light
Lights that plug into the USB port have been very popular and I bought one for use at night.
The first thing I noticed was that it was not too bright, and when one of the LEDs quickly died, it was time to do some upgrading.
Prying the cover apart revealed 3 white LEDs in parallel, with a 30-ohm resistor to limit current to about 60mA.
I want more light, but the 1/10-watt resistor was already putting out major heat.
Step 1: The Upgrade
We know that 3.2-volts is going to the LEDs, and we can measure it with a voltmeter across the LED's wires, but what about the other 1.8-volts supplied by the 5-volt USB?
It's making lots of useless heat in the resistor!
That is what the top circuit shows - of all the power the USB is delivering, a full 36% is being wasted! And 1.8-volt represents 55% of the power going to the LEDs!
That is to say, if we can somehow eliminate the resistor, we'll be able to improve output by 55%! That's like getting an extra 16-miles per gallon from your 30-miles-per-gallon car!
(In EU terms, that's going from 7-l/100km to 5-l/100km!)
The bottom circuit show how we achieve this - we pair each white led with an amber (or yellow) LED.
Amber LEDs, with a Vf of 2-volts, will raise the Vf of the pair to about 5-volts - exactly what the USB delivers!
Step 2: Seeing the Light
With this change, and both circuits drawing about 60mA from the 5-volt USB, the results are startling.
First, with the old design: barely usable.
And then, with the improved design.
Both pictures are taken with the exact exposure and the lights placed an identical distance from the keyboard.
This is the difference 55% makes! Much more usable light, with a warmer (warm-white) glow from the amber LEDs as well.
The 3rd image shows the new stylish "Easter Egg" design, with a less-than-stylish lampshade around it.
With amber LEDs selling on eBay for under 10c each, it's time to upgrade your USB lights!
You can see more of my LED projects here.

Participated in the
Burning Questions: Round 5
5 Comments
14 years ago on Introduction
1/10 watt?? thats gotta be tiny!
15 years ago on Introduction
Did you mesure the current (mA) before and after the modification ? the math are ok, the white leds should be as bright as before (same current), and you would have add 4 ambers leds, which give about half the lumen of the white (a bit less effecient, and 57% of the power of white leds with same current). So you total luminous flux would be multiplied by a factor of 1.5. According to your pics, there is at least a 2 stops differnce, so 4 times the luminus flux. Thought you really gain efficiency (luminous flux multiplied by 1.5 with same total power consumption), i think that the higher luminosity also come from an higher current drawn from the source. Or maybe the digital camera captor is more sensitive to the amber leds than it is to the white ones ? Or maybe the resistor was more than 30 ohms and the usb is providing a little bit more than 5V (which can make a huge change since dI/dV is high for leds) ?
Reply 15 years ago on Introduction
I measured an identical unit (the original was dead already) and it started at 62mA, going down to 55mA. Resistor heating up and expanding, maybe? The mod ran at 57mA. I did not re-use the old LEDs which showed limited reliability, so part of the difference could be just better LEDs. And the photos don't lie - my eyes also see the difference. Same current draw, same conditions, but now I can work comfortably!
Reply 15 years ago on Introduction
I think that the caracteristic of the leds will be more involved in the current VS temperatur variation than carac of the resistor, but temperature is certainly involved. Ok, I missed the fact that you changed the white leds also. It certainly explain part of the result... Anyway, the result on the pics is impressiv, good work !
Reply 15 years ago on Introduction
1.8v * .06A is 108mW, which is already over-limit, and knowing these Chinese manufacturers, what's rated 1/10-watt is seldom that.
So it's no surprise that the little 1/10-watt resistor was running HOT! I measured almost 50oC on my probe.